Status-Quo Bias: Mere Labeling Matters State University

1
Status-Quo Bias Mere Labeling Matters State
University Higher School of Economics June 4,
2009
Maya Bar-Hillel and Avital Moshinsky
2

• The well-known resistance to changing the status
  quo is multiply determined. Some reasons for it
  are completely justified
• It may be better to begin with.
• It may be too costly to change (e.g.,
  transaction costs responsibility costs).

3

• But several behavioral tendencies also serve the
  SQ
• Laziness (mental, too)
• Omission bias, or asymmetric regret
• Routines and habits

4
Decision aversion

• "Chicago were looking for someone to take
  Fermi's place. ... they asked me if I wanted
  to know the salary. 'Oh, no!' I said. "I've
  already decided to stay at Caltech. ... Besides,
  I've decided not to decide any more I'm staying
  at Caltech for good." (Feynman,1985, p. 236).

5

• Also, judgmental errors
• Transaction costs may only seem too high.
• More pertinent to this study The SQ may only
  seem better
• Insofar as these are systematic they are
  sources of bias.

6

• How best to test for a pure SQ bias?
• The status quo is to policies or states of
  affairs as the endowment effect is to goods.
• The endowment effect is most cleanly shown in
  controlled studies where Ss are endowed with a
  good at random.
• Chocolate bar vs. coffee mug example.

7

• Endowing Ss at random with policies is not as
  easy, obviously.
• A common research strategy
• Hypothetical choice among hypothetical states of
  affairs (e.g., Zeckhauser and Samuelson).
• Result A state is more popular when presented
  as the SQ.

8
Another strategy natural experiments (from
Johnson Goldstein, Defaults and donation
decisions Transplantation, 78(12) )
9

• Loss aversion has frequently been evoked to
  explain the endowment effect, as well as the SQ
  bias.
• Thaler, Kahneman Knetsch, 1992
• Baron and Ritov,1994

10
Researchers have also stated that ownership
directly affects valuation.

• Strahilevitz Lowenstein, 1998
• Thaler, Kahneman Knetsch, 1992
• But this has not been tested directly, it has
  just been inferred from the SQs advantage in
  choice.

11

• How could valuation be affected by loss aversion?
  ?
• When comparing two options, A and B, their
  comparative pros and cons may be considered. The
  comparison is reference dependent (i.e., it does
  not necessarily come out the same if carried out
  from a position of having A, or having B, or
  having neither).

12

• E.g., when contemplating an exchange of A for B,
  the advantages of A that would be given up in the
  exchange are losses, because one already had
  them. In contrast, the advantages of B one
  acquires in the exchange constitute gains.

13

• If A is SQ, its pros are potential losses (what
  a pity!), and it cons are potential gains (Good
  riddance).
• If A is NSQ, its pros are potential gains
  (Welcome!), and its cons are potential losses
  (Oy, spare me).
• The example of the exams.

14

• Consequently, A's advantages loom larger when
  considered from the point of view of A versus
  from the point of view of B (since "losses loom
  larger than corresponding gains).
• Whereas the reverse is true for As disadvantages
  (or for Bs advantages).

15

• So we should be able to see a status quo bias
  even without any choice or decision. Just being
  the SQ should enhance an options attractiveness.
  
• We call this hypothesized effect
• Status Quo Label Bias -- SQLB.

16
Status Quo Label Bias

• Definition
• The difference between the attractiveness of some
  alternative when it is the SQ and when it is not.
  

17

• Some caveats
• SQLB does not predict that the status quo will
  always be liked more than an alternative only
  that an alternative will be liked more when it is
  the status quo.

18

• SQLB can occur even when the status quo is not
  preferred to some alternative it merely
  contributes to a status quo bias.
• SQLB is not a bias of choice or decision, it is
  bias in judgment (whereas the SQB is a bias of
  choice).

19
A double pan balance metaphor

• In a perfect pan balance, the order in which
  objects are placed on the pans does not matter.
  But if the fulcrum is rusty, the second pan needs
  to overcome the weight on first pan and the
  friction. This is SQ bias in choice.
• If adding a tiny sticker that says SQ lowers
  the pan which holds it that is the SQLB.

20

• The present study shows
• There is an SQLB policies are liked better when
  they are accompanied by the SQ label.
• The magnitude of the SQLB is predictable from
  that of loss aversion.
• SQLB occurs even when alternative mechanisms are
  obviated -- but not when loss aversion is
  blocked.

21
The study.

• To mimic the endowment effect studies, we
  endowed Ss with policies at random.
• How? By choosing issues on which we assumed (and
  verified) that Ss did not know the true state of
  affairs.

22

• Participants and Procedure.
• Respondents About 900 undergraduate students
  from The Hebrew University (53 female most 21 -
  25 years old).
• They were approached after lectures in their
  lecture rooms and asked to stay behind and answer
  a short questionnaire. In exchange, one
  respondent (determined by lottery), would win a
  monetary prize.
• Prizes (in NIS) were about 2-times-N (rounded
  up), where N was the number of Ss remaining in
  the lecture hall. The average prize was about
  100 NIS (25).
• Students were assigned at random to
  questionnaires, and queried about a single policy
  issue.
• The questionnaire rarely took more than 10
  minutes to complete.

23

• Tasks.
• The questionnaires concerned 2 alternative
  proposals on some issue. Tasks, in order, were
  
• Think of and list the pros and cons of either the
  SQ policy or of its alternative (but not of
  both), and to rate the importance of each
  consideration they listed on a scale from 1 ("not
  so important") to 4 ("very important").
• 2. Which policy do you consider to be better?
• Rate each policy separately, on a scale of 1
  ("very bad") to 6 ("very good").
• (On a later page) Ss were asked whether they knew
  which the prevailing policy really was, and then
  debriefed.

24
No decision making was involved, either
hypothetical or real. Just evaluation.

• Is there still a status quo label bias?
• (In the following, A will denote the policy
  which, as SQ, Ss favored more).

25
B prevails A prevails Percent of respondents who prefer A (listed) to B, when
Advertising alcohol (restricted)
Arts and crafts (7 vs. 5) in school (5 vs. 3)
Affirmative action in HU (yes)
Feeding alley cats (allowed)
Prostitution (with restrictions)
Rescuing people (mandatory)
Rottweilers (mostly prohibited)
Child testimony, sex crimes (not like adults)
Statute of limitations, civil (only 2-tier)
26
B prevails A prevails Percent of respondents who prefer A (listed) to B, when
55 92 Advertising alcohol (restricted)
57 66 87 86 Arts and crafts (7 vs. 5) in school (5 vs. 3)
63 83 Affirmative action in HU (yes)
59 81 Feeding alley cats (allowed)
56 71 Prostitution (with restrictions)
42 70 Rescuing people (mandatory)
55 67 Rottweilers (mostly prohibited)
62 62 Child testimony, sex crimes (not like adults)
56 59 Statute of limitations, civil (only 2-tier)
27
Percent of (almost 900) respondents who think
policy A is better, when it is labeled SQ (left
bars), and when it is not (right bars).
28

• The overall effect was 73 favoring the favorite
  policy when it was labeled SQ compared to 56
  favoring it when it was not.
• A similar picture emerges with the attractiveness
  ratings.

29
B SQA SQB A (listed) SQA SQB Ratings of policies
2.1 3.5 4.7 3.7 Advertising alcohol (restricted)
2.5 4.0 2.4 3.8 4.0 3.9 4.10 4.14 Arts and crafts (7 vs. 5) in school (5 vs. 3)
2.6 3.6 4.0 3.7 Affirmative action (yes)
2.6 3.4 4.3 3.5 Feeding alley cats (allowed)
2.9 3.4 3.53 3.51 Prostitution (with restrictions)
3.3 3.9 4.1 3.3 Rescuing people (mandatory)
3.1 3.0 3.7 3.8 Rottweilers (prohibited)
3.2 3.4 3.90 3.85 Child testimony (not like adults)
3.3 3.4 3.7 3.8 Statute of limitations (more uniform)
30

• Overall, there were 16 successes in 20
  opportunities.
• On average A policies rated 4 as SQ and only 3.7
  as NQS B polices rated 3.5 as SQ and only 2.9
  as NSQ (both highly significant).
• Are there alternative explanations?

31

• The mere ownership effect.
• Beggan, 1992
• I like it because it is mine.
• System justification theory
• Jost and colleagues
• What is is right

32

• These are real effects.
• But do they explain the present effect?
• We tested whether the actual template used
  triggers a presumption that SQ is better.

33

• 22 students were given the following question
• Suppose you are told that "The prevailing
  policy in Israel is such-and-such sic."
• In addition you are told that "A suggestion has
  been made to change it to so-and-so".
• What does this formulation make you think?
• - The prevailing policy is the better one.
  2
• - The proposed alternative is the better one.
  20

34

• For another 29 Israeli students, after the same
  opening sentences, the question was replaced by
• Which of the following inferences seems to you
  more reasonable?
• - The prevailing policy is probably better,
  because those who decided in its favor must have
  had their reasons. 3
• The prevailing policy is probably not so good,
  because people are considering replacing it.
  
• 26

35
33 others were asked instead

• - Which of the following inferences seems to
  you more reasonable?
• The prevailing policy is probably good,
  because those who decided in
• its favor must have had their reasons, which
  can be trusted. (14)
• The prevailing policy is probably not so
  good, because those who decided
• in its favor must have had their reasons, which
  cannot be trusted. (19)

36
Similar (if somewhat attenuated) results were
obtained even when the phrase about a suggested
change was omitted. 25 students were asked

• Suppose you are told that "The prevailing
  policy in Israel is such-and-such."
• - What does this formulation make you think?
• The prevailing policy is better than some
  alternative policy. ( 8)
• The prevailing policy is worse than some
  alternative policy. (17)

37

• Finally, 109 Ss were told, in the very same words
  as in the study described "A certain policy
  prevails in Israel regarding the advertising of
  alcohol on TV. A suggestion has been made to
  introduce some changes in it".
• In addition to not saying who raised the
  challenge, this formulation does not say what the
  prevailing policy is, nor what the proposed
  changes are. Since one can hardly list the pros
  and cons of a policy that is not specified, no
  such list was requested.

38

• Respondents were asked to state which policy they
  judged to be better (task 2), and to rate each of
  the 2 policies on a scale of 1 to 6, as before
  (task 3).
• This was done with the 3 issues that showed the
  largest SQLB (alcohol, arts crafts, mandatory
  rescue), and for a 4th, generic, issue.

39
N Rating of NSQ Rating of SQ Number choosing NSQ Number choosing SQ Policy
30 3.4 3.8 11 19 Advertising alcohol
28 3.9 3.8 15 13 Arts crafts in school
28 3.6 3.8 14 14 Rescuing people in peril
23 3.7 3.5 12 11 "Such-and-such"
109 3.6 3.7 48 52 Total
40
The same was found even with no mention of a
challenge.
N Rating of NSQ Rating of SQ Number choosing NSQ Number choosing SQ Policy
33 3.7 3.8 15 18 Advertising alcohol
28 4.2 3.8 20 11 Arts crafts in school
28 3.6 3.8 14 21 Rescuing people in peril
36 4.3 3.2 25 9 "Such-and-such"
109 4.0 3.7 56 44 Total
41

• We can reject presumption in favor of the SQ.
• There either is the reverse presumption or
  none.
• This, in spite of preserving all the social
  cues.

42
Why were results different in the actual study
(as shown in Tables 1 and 2)?

• Because the non-specific template provided no
  grist for the loss-aversion mill.
• And the SQLB as we hypothesized it results from
  loss aversion -- so may not appear where there is
  no loss aversion.

43

• A policys attractiveness is related to the
  balance of its pros and cons.
• The policies attractiveness ratings were
  positively correlated with their net (weighted)
  balance of pros and cons.
• Pearsons r0.65, when policies were the units.
• Pearsons r0.52, when participants were the
  units.
• Both highly significant

44

• Recall
• Losses are pros of the SQ,
• and cons of the NSQ
• Gains are cons of the SQ,
• and pros of the NSQ

45

• We weighted each listed pro or con by the
  importance weight given it by the participant,
  and calculated losses and gains as in the
  previous slide.

46

• The correlation between the magnitude of the SQLB
  and that of loss aversion was 0.59 for the
  popularity measure, and 0.72 for the
  attractiveness measure.
• The magnitude of the SQLB can be predicted from
  the magnitude of loss aversion.

47

• Thank you for your attention!